On 12/26/2010 11:29 PM, Hu Tao wrote:
Do nothing if caller passes NULL as jobdata to virThreadPoolSendJob(). --- src/util/threadpool.c | 3 +++ src/util/threadpool.h | 1 - 2 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/src/util/threadpool.c b/src/util/threadpool.c index 1213862..2907478 100644 --- a/src/util/threadpool.c +++ b/src/util/threadpool.c @@ -192,6 +192,9 @@ int virThreadPoolSendJob(virThreadPoolPtr pool, { virThreadPoolJobPtr job;
+ if (!jobData) + return -1; +
This hunk is wrong. jobData is opaque - we cannot assume anything about it's meaning. A job may very well have something to do even with a NULL job data, and refusing to run a job just because one of its two opaque arguments is NULL is no better than marking the parameter as ATTRIBUTE_NONNULL when registering the job.
+++ b/src/util/threadpool.h @@ -42,7 +42,6 @@ void virThreadPoolFree(virThreadPoolPtr pool);
int virThreadPoolSendJob(virThreadPoolPtr pool, void *jobdata) ATTRIBUTE_NONNULL(1) - ATTRIBUTE_NONNULL(2) ATTRIBUTE_RETURN_CHECK;
But this hunk is okay. I've applied just this hunk, as it is technically a bug fix (over-strict compilation markings), and minimal impact for inclusion in 0.8.7. -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org